Do Magnet Schools Increase Segregation in Nearby Public and Private Schools?
Yoon (2026) examines how opening interdistrict magnet schools affects enrollment and student demographics in nearby public and private schools. She analyzes Connecticut school enrollment and demographic data from 1998 to 2020, using a difference-in-differences approach to estimate causal effects. She finds that magnet school openings reduced enrollment by 6.1% in nearby private K–8 schools and 5.3% in traditional public high schools. The proportion of White students in private K–8 schools declined by 2.2 percentage points, while the share of economically disadvantaged students in public high schools increased by 7.7 percentage points, suggesting that more affluent families disproportionately leave.
Why This Article Was Selected for The Policy Scientist
School choice policies reshape educational opportunities, enrollment patterns, and the distribution of students across schools. As states continue expanding alternatives to traditional public education, understanding the consequences for schools that lose students is increasingly important. Policies intended to reduce segregation may also change the composition of neighboring schools, creating broader implications for educational inequality. Published in Education Finance and Policy, a respected journal in education policy research, this study makes an important contribution by examining these unintended effects. Its comprehensive longitudinal data and credible difference-in-differences design strengthen causal inference.
Full Citation and Link to Article
Yoon, Y. (2026). Who leaves? Interdistrict magnet school openings and enrollment dynamics in nearby schools. Education Finance and Policy. Advance online publication.
Central Research Question
Does opening a new magnet school affect enrollment and the racial and socioeconomic composition of nearby traditional public and private schools? More specifically, do magnet schools designed to reduce segregation across school district boundaries inadvertently increase segregation among students who remain in surrounding schools?
Connecticut provides an important setting for investigating these questions. Following the landmark Sheff v. O’Neill school desegregation litigation, the state expanded interdistrict magnet schools to bring students from different communities into racially and economically integrated educational environments. Unlike conventional magnet schools, these institutions recruit students across district boundaries and offer specialized curricula without charging families tuition. However, their expansion may also encourage families to leave existing schools, potentially changing the characteristics of the students who remain. The study examines whether these consequences differ across public and private schools, elementary and secondary education, and communities subject to court-ordered desegregation.
Previous Literature
Previous research establishes that magnet schools can increase racial integration among participating students, although their effects on segregation across entire school systems remain uncertain. Bifulco et al. (2009a) found that Connecticut's interdistrict magnet schools provided students from highly segregated urban communities with access to more integrated educational environments. However, Saporito (2003) demonstrated that White and economically advantaged families were disproportionately likely to apply to magnet schools in Philadelphia, raising questions about how voluntary school choice affects the composition of neighborhood schools. Kho et al. (2022) similarly identified evidence that magnet schools in Tennessee and North Carolina attract relatively high-achieving and economically advantaged students.
Research on charter schools provides additional evidence about the consequences of expanding educational choice. Alcaino and Jennings (2020) found that charter school growth contributed to racial and socioeconomic segregation, while Monarrez et al. (2022) documented more complicated effects, including increased segregation in some settings and reduced segregation between districts in others. Research also suggests that public school alternatives can reduce private school enrollment. Waddington (2012) found that charter school expansion accelerated enrollment declines in nearby Catholic schools, while Dougherty et al. (2025) documented similar effects nationally. The present study extends this literature by examining the consequences of interdistrict magnet school openings for surrounding institutions rather than focusing exclusively on students attending the new schools.
Data
The study uses the National Longitudinal School Database, which combines information from the federal Common Core of Data and Private School Survey. The analysis covers Connecticut schools from the 1998–1999 through 2019–2020 academic years, providing approximately two decades of information about school enrollment, racial composition, geographic location, and institutional characteristics. Public school observations are available annually, while private school information is collected every two years.
The researcher identifies 84 full-time interdistrict magnet schools and links their opening dates and geographic locations to surrounding schools. The resulting dataset includes 949 traditional public K–8 schools, 454 traditional public high schools, 278 private K–8 schools, and 128 private high schools. Outcomes include total enrollment, the percentage of students identifying as White, Black, or Hispanic, and the percentage eligible for free or reduced-price lunch, which serves as an indicator of economic disadvantage. The analysis also distinguishes Catholic from non-Catholic private schools and schools located within Hartford County, where court-ordered desegregation policies were particularly consequential.
Methods
The researcher employs a staggered difference-in-differences design to estimate the effects of new magnet school openings. Because magnet schools opened in different communities at different times, the analysis compares changes in enrollment and student composition at schools located near newly opened magnet schools with changes at schools that had not yet experienced such an opening.
The primary treatment is the opening of an interdistrict magnet school within 2.5 miles of an existing school serving overlapping grade levels. Rather than relying exclusively on conventional two-way fixed-effects regression, the study applies the Callaway and Sant’Anna (2021) estimator, which accommodates differences in treatment timing and allows treatment effects to vary across schools and years. Event-study models examine changes before and after openings, while additional analyses consider alternative distance thresholds, estimation methods, and geographic characteristics.
This approach strengthens causal interpretation by comparing enrollment changes across differently exposed schools over time. Nevertheless, magnet school locations and opening dates were not randomly assigned. Consequently, the findings depend on assumptions about how enrollment would have evolved without the openings, particularly the absence of unobserved changes that simultaneously influence magnet school placement and surrounding enrollment patterns.
Findings/Size Effects
Magnet school openings produced measurable enrollment declines in nearby private elementary and middle schools and traditional public high schools. However, the consequences differed substantially across school sectors and grade levels.
For private K–8 schools, opening a magnet school within 2.5 miles reduced enrollment by approximately 12.5 students, representing a 6.1% decline relative to average enrollment of 204 students. Traditional public high schools experienced an average reduction of 64.5 students, equivalent to a 5.3% decline relative to their baseline enrollment of approximately 1,214 students. By contrast, the study found no statistically significant enrollment changes in traditional public K–8 schools or private high schools.
The demographic consequences were particularly notable. In nearby private K–8 schools, the proportion of White students declined by 2.2 percentage points, from a baseline of approximately 62%. This represents a 3.5% relative reduction in the White student share. Consequently, Black and Hispanic students remaining in these institutions experienced reduced exposure to White classmates. These changes were concentrated primarily in Catholic schools, suggesting that families attending relatively affordable religious schools may respond differently to newly available public alternatives.
Patterns differed among non-Catholic private K–8 schools. Two years after nearby magnet schools opened, the proportion of Black students declined by approximately 2.5 percentage points, while the Hispanic share declined by 1.8 percentage points. The White student share increased, indicating that demographic responses to magnet school expansion differed across private school categories.
Traditional public high schools experienced especially pronounced changes in socioeconomic composition. Following nearby magnet school openings, the percentage of students eligible for free or reduced-price lunch increased by approximately 7.7 percentage points, from a baseline of 42% to nearly 50%. This represents an 18.3% relative increase in the proportion of economically disadvantaged students.
These results suggest that students from relatively higher-income families were disproportionately likely to leave traditional public high schools when new magnet school options became available. The effects were particularly pronounced in the Hartford region, where interdistrict magnet schools were established partly to satisfy court-ordered desegregation requirements. Importantly, the researcher did not identify corresponding statistically significant changes in the racial composition of traditional public high schools.
Additional analyses using alternative distance thresholds, statistical estimators, and sample restrictions generally supported the principal findings. Placebo tests examining magnet school openings serving nonoverlapping grade levels also produced no significant effects, strengthening the interpretation that observed enrollment changes reflected competition among schools serving similar students rather than general neighborhood trends.
Conclusion
The study demonstrates that expanding magnet schools can change enrollment patterns and student demographics beyond the institutions directly participating in school choice programs. Although Connecticut's interdistrict magnet schools were established to reduce racial and socioeconomic isolation, their openings also altered the composition of nearby schools, particularly private K–8 institutions and traditional public high schools.
The findings highlight an important distinction between improving integration within individual magnet schools and reducing segregation throughout an educational system. A policy can increase opportunities for students attending newly established schools while simultaneously changing the educational environments of students who remain elsewhere.
These results suggest that evaluations of school choice policies should examine the entire local education market rather than concentrating exclusively on participating students. Nevertheless, Connecticut's distinctive legal framework, extensive interdistrict magnet system, and geographic characteristics may limit the applicability of the findings to other states. Furthermore, the analysis documents changes in school-level enrollment and composition rather than directly tracking individual student transfers or measuring subsequent academic achievement. Future research could investigate these additional outcomes to determine whether the demographic changes identified here translate into measurable differences in educational experiences and student performance.
Comments